Independent research prototype · v0.3.0

A knowledge graph of the psychedelic experience

Relating self-reported subjective experience to molecular structure and receptor pharmacology, and modelling the structure-activity relationships that connect them.

chemistry → pharmacology → phenomenology → outcomes

What this is

An individual report is an anecdote. Aggregated across many reports and joined to each molecule's receptor-binding profile, the same material becomes analysable.

PsycheGraph represents each experience as a reified event, linked to the substances taken, to the effects described (each with an intensity, a time of onset, and a verbatim supporting quotation), and to any lasting or adverse outcomes. A parallel molecular layer records structure and receptor affinities, so reported phenomenology can be examined against receptor pharmacology rather than discussed in isolation.

The current corpus is Erowid, chosen as an openly available basis for a proof of concept. The schema is designed to accommodate more rigorously characterised sources as the work matures, including Shulgin's reports and structured psychometric instruments.

Chemistry

Structure

SMILES and InChI, chemical class, prodrug relationships (PubChem).

Pharmacology

Receptors

Binding affinities (Kᵢ) across serotonin, dopamine, NMDA, and opioid targets (PDSP).

Phenomenology

Experience

Effects recorded as timed, quoted events within a hierarchical taxonomy.

Outcomes

Therapeutic and adverse

Self-reported lasting effects and adverse events.

Current status

An early-stage prototype. The figures below are read from the running graph at page load.

Corpus figures withheld

Report, substance and word counts are derived from Erowid experience reports and are withheld pending a data-rights agreement. The data is retained and the analysis runs; only its publication waits. The enrichment coverage below is measured against PubChem and PDSP, which carry no such restriction.

38
substances deep-characterised
100%
with molecular structure
57.9%
with receptor data

The class of question it is designed to answer

Illustrative rather than a result. The intent is to show the form of the query.

-- Which phenomena covary with high 5-HT2A affinity (Kᵢ < 100 nM)?
phenomenon                  mean_intensity   n_substances   confidence
transpersonal.unity              0.81             n           CI, low-n flagged
perceptual.visual.geometry       0.74             n           CI, low-n flagged
-- joined:  substance -> receptor_binding -> receptors
--          substance <- experience event -> phenomenon
Every such statement is reported with a sample count, a confidence interval, and a low-sample flag. The objective is hypothesis generation under explicit uncertainty, not efficacy claims.

Method and limitations

The evidence is self-reported, and the design treats that as a first-order constraint.

Experience reports carry well-documented biases. Reporting is subject to selection effects; drug identity and dose are typically unverified; polydrug use confounds attribution; timing is inconsistent; vocabulary is culturally primed; and outcomes are self-assessed rather than clinically measured. The response is statistical: per-statistic sample counts, confidence intervals, low-sample flags, and complete provenance, with every extracted fact linked to the quotation from which it was derived.

The approach has precedent. Ballentine, Friedman and Bzdok (Science Advances, 2022) applied natural-language processing to a corpus of Erowid reports across 40 receptor subtypes, and Zamberlan et al. (2018) reported that binding-affinity similarity tracks the semantic similarity of reports. PsycheGraph seeks to generalise that work into a reusable, provenance-tracked graph spanning multiple corpora, with the molecular join treated as first-class.

The constraint that keeps the work honest. Two risks shape the design. Ineffability is itself a reported feature, since the most profound states may be the least describable; sparse regions of the data are therefore not read as an absence of experience. Circularity is the deeper problem: a space defined from reports and validated against reports would characterise language rather than experience. The necessary constraint comes from outside the text, in receptor pharmacology, assayed identity, and dose-response. This is why the structure-activity layer is epistemically load-bearing rather than a convenience.

Data and provenance

Erowid Experience Vaults

The example corpus of first-person reports. Attribution is preserved and the material is used for research.

PubChem

Molecular structure: SMILES, InChI, formula, and weight.

PDSP Kᵢ Database

Receptor binding affinities (UNC and NIMH PDSP).

Planned

Shulgin (PIHKAL and TIHKAL), psychometric instruments (5D-ASC, MEQ), and further corpora.

Research trajectory

Staged and labelled: falsifiable questions first, a longer research programme behind them.

The near-term work strengthens the receptor-to-experience mapping as the corpus grows, and holds it to out-of-sample tests rather than in-sample fit. An active research programme connects the graph to mechanistic models of psychedelic action.

Claims are held to the evidence at every stage. Methods, preregistrations, and the programme's open questions are shared with collaborators rather than announced ahead of results.

Collaboration and feedback

PsycheGraph is an independent research project. Critical feedback from researchers is welcome. I would be glad to hear from anyone who regards the method as useful or as flawed, who is open to collaboration or to validation against real data, or who can point to corpora worth incorporating.

Email [email protected]